dlls/msvcrt/mbcs.c

   1 /*
   2  * msvcrt.dll mbcs functions
   3  *
   4  * Copyright 1999 Alexandre Julliard
   5  * Copyright 2000 Jon Griffths
   6  *
   7  * This library is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Lesser General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2.1 of the License, or (at your option) any later version.
  11  *
  12  * This library is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Lesser General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser General Public
  18  * License along with this library; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
  20  *
  21  * FIXME
  22  * Not currently binary compatible with win32. MSVCRT_mbctype must be
  23  * populated correctly and the ismb* functions should reference it.
  24  */
  25
  26 #include <stdio.h>
  27
  28 #include "msvcrt.h"
  29 #include "wine/unicode.h"
  30 #include "wine/debug.h"
  31
  32 WINE_DEFAULT_DEBUG_CHANNEL(msvcrt);
  33
  34 unsigned char MSVCRT_mbctype[257] = { 0 };
  35
  36 /* It seems that the data about valid trail bytes is not available from kernel32
  37  * so we have to store is here. The format is the same as for lead bytes in CPINFO */
  38 struct cp_extra_info_t
  39 {
  40     int cp;
  41     BYTE TrailBytes[MAX_LEADBYTES];
  42 };
  43
  44 static struct cp_extra_info_t g_cpextrainfo[] =
  45 {
  46     {932, {0x40, 0x7e, 0x80, 0xfc, 0, 0}},
  47     {936, {0x40, 0xfe, 0, 0}},
  48     {949, {0x41, 0xfe, 0, 0}},
  49     {950, {0x40, 0x7e, 0xa1, 0xfe, 0, 0}},
  50     {1361, {0x31, 0x7e, 0x81, 0xfe, 0, 0}},
  51     {20932, {1, 255, 0, 0}},  /* seems to give different results on different systems */
  52     {0, {1, 255, 0, 0}}       /* match all with FIXME */
  53 };
  54
  55 /* Maps cp932 single byte character to multi byte character */
  56 static const unsigned char mbbtombc_932[] = {
  57   0x40,0x49,0x68,0x94,0x90,0x93,0x95,0x66,0x69,0x6a,0x96,0x7b,0x43,0x7c,0x44,0x5e,
  58   0x4f,0x50,0x51,0x52,0x53,0x54,0x55,0x56,0x57,0x58,0x46,0x47,0x83,0x81,0x84,0x48,
  59   0x97,0x60,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x6b,0x6c,0x6d,0x6e,
  60   0x6f,0x70,0x71,0x72,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x6d,0x8f,0x6e,0x4f,0x76,
  61   0x77,0x78,0x79,0x6d,0x8f,0x6e,0x4f,0x51,0x65,0x81,0x82,0x83,0x84,0x85,0x86,0x87,
  62   0x88,0x89,0x8a,0x8b,0x8c,0x8d,0x8e,0x8f,0x90,0x91,0x92,0x93,0x94,0x95,0x50,
  63        0x42,0x75,0x76,0x41,0x45,0x92,0x40,0x42,0x44,0x46,0x48,0x83,0x85,0x87,0x62,
  64   0x5b,0x41,0x43,0x45,0x47,0x49,0x4a,0x4c,0x4e,0x50,0x52,0x54,0x56,0x58,0x5a,0x5c,
  65   0x5e,0x60,0x63,0x65,0x67,0x69,0x6a,0x6b,0x6c,0x6d,0x6e,0x71,0x74,0x77,0x7a,0x7d,
  66   0x7e,0x80,0x81,0x82,0x84,0x86,0x88,0x89,0x8a,0x8b,0x8c,0x8d,0x8f,0x93,0x4a,0x4b };
  67
  68 /* Maps multibyte cp932 punctuation marks to single byte equivalents */
  69 static const unsigned char mbctombb_932_punct[] = {
  70   0x20,0xa4,0xa1,0x2c,0x2e,0xa5,0x3a,0x3b,0x3f,0x21,0xde,0xdf,0x00,0x00,0x00,0x5e,
  71   0x7e,0x5f,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0xb0,0x00,0x00,0x2f,0x00,
  72   0x00,0x00,0x7c,0x00,0x00,0x60,0x27,0x00,0x22,0x28,0x29,0x00,0x00,0x5b,0x5d,0x7b,
  73   0x7d,0x00,0x00,0x00,0x00,0xa2,0xa3,0x00,0x00,0x00,0x00,0x2b,0x2d,0x00,0x00,0x00,
  74   0x00,0x3d,0x00,0x3c,0x3e,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x5c,
  75   0x24,0x00,0x00,0x25,0x23,0x26,0x2a,0x40};
  76
  77 /* Maps multibyte cp932 hiragana/katakana to single-byte equivalents */
  78 static const unsigned char mbctombb_932_kana[] = {
  79   0xa7,0xb1,0xa8,0xb2,0xa9,0xb3,0xaa,0xb4,0xab,0xb5,0xb6,0xb6,0xb7,0xb7,0xb8,0xb8,
  80   0xb9,0xb9,0xba,0xba,0xbb,0xbb,0xbc,0xbc,0xbd,0xbd,0xbe,0xbe,0xbf,0xbf,0xc0,0xc0,
  81   0xc1,0xc1,0xaf,0xc2,0xc2,0xc3,0xc3,0xc4,0xc4,0xc5,0xc6,0xc7,0xc8,0xc9,0xca,0xca,
  82   0xca,0xcb,0xcb,0xcb,0xcc,0xcc,0xcc,0xcd,0xcd,0xcd,0xce,0xce,0xce,0xcf,0xd0,0xd1,
  83   0xd2,0xd3,0xac,0xd4,0xad,0xd5,0xae,0xd6,0xd7,0xd8,0xd9,0xda,0xdb,0xdc,0xdc,0xb2,
  84   0xb4,0xa6,0xdd,0xb3,0xb6,0xb9};
  85
  86 static MSVCRT_wchar_t msvcrt_mbc_to_wc(unsigned int ch)
  87 {
  88   MSVCRT_wchar_t chW;
  89   char mbch[2];
  90   int n_chars;
  91
  92   if (ch <= 0xff) {
  93     mbch[0] = ch;
  94     n_chars = 1;
  95   } else {
  96     mbch[0] = (ch >> 8) & 0xff;
  97     mbch[1] = ch & 0xff;
  98     n_chars = 2;
  99   }
 100   if (!MultiByteToWideChar(get_mbcinfo()->mbcodepage, 0, mbch, n_chars, &chW, 1))
 101   {
 102     WARN("MultiByteToWideChar failed on %x\n", ch);
 103     return 0;
 104   }
 105   return chW;
 106 }
 107
 108 static inline MSVCRT_size_t u_strlen( const unsigned char *str )
 109 {
 110   return strlen( (const char*) str );
 111 }
 112
 113 static inline unsigned char* u_strncat( unsigned char* dst, const unsigned char* src, MSVCRT_size_t len )
 114 {
 115   return (unsigned char*)strncat( (char*)dst, (const char*)src, len);
 116 }
 117
 118 static inline int u_strcmp( const unsigned char *s1, const unsigned char *s2 )
 119 {
 120   return strcmp( (const char*)s1, (const char*)s2 );
 121 }
 122
 123 static inline int u_strcasecmp( const unsigned char *s1, const unsigned char *s2 )
 124 {
 125   return strcasecmp( (const char*)s1, (const char*)s2 );
 126 }
 127
 128 static inline int u_strncmp( const unsigned char *s1, const unsigned char *s2, MSVCRT_size_t len )
 129 {
 130   return strncmp( (const char*)s1, (const char*)s2, len );
 131 }
 132
 133 static inline int u_strncasecmp( const unsigned char *s1, const unsigned char *s2, MSVCRT_size_t len )
 134 {
 135   return strncasecmp( (const char*)s1, (const char*)s2, len );
 136 }
 137
 138 static inline unsigned char *u_strchr( const unsigned char *s, unsigned char x )
 139 {
 140   return (unsigned char*) strchr( (const char*)s, x );
 141 }
 142
 143 static inline unsigned char *u_strrchr( const unsigned char *s, unsigned char x )
 144 {
 145   return (unsigned char*) strrchr( (const char*)s, x );
 146 }
 147
 148 static inline unsigned char *u_strtok( unsigned char *s, const unsigned char *delim )
 149 {
 150   return (unsigned char*) strtok( (char*)s, (const char*)delim );
 151 }
 152
 153 static inline unsigned char *u__strset( unsigned char *s, unsigned char c )
 154 {
 155   return (unsigned char*) _strset( (char*)s, c);
 156 }
 157
 158 static inline unsigned char *u__strnset( unsigned char *s, unsigned char c, MSVCRT_size_t len )
 159 {
 160   return (unsigned char*) MSVCRT__strnset( (char*)s, c, len );
 161 }
 162
 163 static inline MSVCRT_size_t u_strcspn( const unsigned char *s, const unsigned char *rej )
 164 {
 165   return strcspn( (const char *)s, (const char*)rej );
 166 }
 167
 168 /*********************************************************************
 169  *              __p__mbctype (MSVCRT.@)
 170  */
 171 unsigned char* CDECL __p__mbctype(void)
 172 {
 173   return get_mbcinfo()->mbctype;
 174 }
 175
 176 /*********************************************************************
 177  *              __p___mb_cur_max(MSVCRT.@)
 178  */
 179 int* CDECL __p___mb_cur_max(void)
 180 {
 181   return &get_locinfo()->mb_cur_max;
 182 }
 183
 184 /*********************************************************************
 185  *              ___mb_cur_max_func(MSVCRT.@)
 186  */
 187 int CDECL MSVCRT____mb_cur_max_func(void)
 188 {
 189   return get_locinfo()->mb_cur_max;
 190 }
 191
 192 /* ___mb_cur_max_l_func - not exported in native msvcrt */
 193 int* CDECL ___mb_cur_max_l_func(MSVCRT__locale_t locale)
 194 {
 195   MSVCRT_pthreadlocinfo locinfo;
 196
 197   if(!locale)
 198     locinfo = get_locinfo();
 199   else
 200     locinfo = locale->locinfo;
 201
 202   return &locinfo->mb_cur_max;
 203 }
 204
 205 /*********************************************************************
 206  * INTERNAL: _setmbcp_l
 207  */
 208 int _setmbcp_l(int cp, LCID lcid, MSVCRT_pthreadmbcinfo mbcinfo)
 209 {
 210   const char format[] = ".%d";
 211
 212   int newcp;
 213   CPINFO cpi;
 214   BYTE *bytes;
 215   WORD chartypes[256];
 216   char bufA[256];
 217   WCHAR bufW[256];
 218   int charcount;
 219   int ret;
 220   int i;
 221
 222   if(!mbcinfo)
 223       mbcinfo = get_mbcinfo();
 224
 225   switch (cp)
 226   {
 227     case _MB_CP_ANSI:
 228       newcp = GetACP();
 229       break;
 230     case _MB_CP_OEM:
 231       newcp = GetOEMCP();
 232       break;
 233     case _MB_CP_LOCALE:
 234       newcp = get_locinfo()->lc_codepage;
 235       if(newcp)
 236           break;
 237       /* fall through (C locale) */
 238     case _MB_CP_SBCS:
 239       newcp = 20127;   /* ASCII */
 240       break;
 241     default:
 242       newcp = cp;
 243       break;
 244   }
 245
 246   if(lcid == -1) {
 247     sprintf(bufA, format, newcp);
 248     mbcinfo->mblcid = MSVCRT_locale_to_LCID(bufA);
 249   } else {
 250     mbcinfo->mblcid = lcid;
 251   }
 252
 253   if(mbcinfo->mblcid == -1)
 254   {
 255     WARN("Can't assign LCID to codepage (%d)\n", mbcinfo->mblcid);
 256     mbcinfo->mblcid = 0;
 257   }
 258
 259   if (!GetCPInfo(newcp, &cpi))
 260   {
 261     WARN("Codepage %d not found\n", newcp);
 262     *MSVCRT__errno() = MSVCRT_EINVAL;
 263     return -1;
 264   }
 265
 266   /* setup the _mbctype */
 267   memset(mbcinfo->mbctype, 0, sizeof(unsigned char[257]));
 268   memset(mbcinfo->mbcasemap, 0, sizeof(unsigned char[256]));
 269
 270   bytes = cpi.LeadByte;
 271   while (bytes[0] || bytes[1])
 272   {
 273     for (i = bytes[0]; i <= bytes[1]; i++)
 274       mbcinfo->mbctype[i + 1] |= _M1;
 275     bytes += 2;
 276   }
 277
 278   if (cpi.MaxCharSize > 1)
 279   {
 280     /* trail bytes not available through kernel32 but stored in a structure in msvcrt */
 281     struct cp_extra_info_t *cpextra = g_cpextrainfo;
 282
 283     mbcinfo->ismbcodepage = 1;
 284     while (TRUE)
 285     {
 286       if (cpextra->cp == 0 || cpextra->cp == newcp)
 287       {
 288         if (cpextra->cp == 0)
 289           FIXME("trail bytes data not available for DBCS codepage %d - assuming all bytes\n", newcp);
 290
 291         bytes = cpextra->TrailBytes;
 292         while (bytes[0] || bytes[1])
 293         {
 294           for (i = bytes[0]; i <= bytes[1]; i++)
 295             mbcinfo->mbctype[i + 1] |= _M2;
 296           bytes += 2;
 297         }
 298         break;
 299       }
 300       cpextra++;
 301     }
 302   }
 303   else
 304     mbcinfo->ismbcodepage = 0;
 305
 306   /* we can't use GetStringTypeA directly because we don't have a locale - only a code page
 307    */
 308   charcount = 0;
 309   for (i = 0; i < 256; i++)
 310     if (!(mbcinfo->mbctype[i + 1] & _M1))
 311       bufA[charcount++] = i;
 312
 313   ret = MultiByteToWideChar(newcp, 0, bufA, charcount, bufW, charcount);
 314   if (ret != charcount)
 315     ERR("MultiByteToWideChar of chars failed for cp %d, ret=%d (exp %d), error=%d\n", newcp, ret, charcount, GetLastError());
 316
 317   GetStringTypeW(CT_CTYPE1, bufW, charcount, chartypes);
 318
 319   charcount = 0;
 320   for (i = 0; i < 256; i++)
 321     if (!(mbcinfo->mbctype[i + 1] & _M1))
 322     {
 323       if (chartypes[charcount] & C1_UPPER)
 324       {
 325         mbcinfo->mbctype[i + 1] |= _SBUP;
 326         bufW[charcount] = tolowerW(bufW[charcount]);
 327       }
 328       else if (chartypes[charcount] & C1_LOWER)
 329       {
 330         mbcinfo->mbctype[i + 1] |= _SBLOW;
 331         bufW[charcount] = toupperW(bufW[charcount]);
 332       }
 333       charcount++;
 334     }
 335
 336   ret = WideCharToMultiByte(newcp, 0, bufW, charcount, bufA, charcount, NULL, NULL);
 337   if (ret != charcount)
 338     ERR("WideCharToMultiByte failed for cp %d, ret=%d (exp %d), error=%d\n", newcp, ret, charcount, GetLastError());
 339
 340   charcount = 0;
 341   for (i = 0; i < 256; i++)
 342   {
 343     if(!(mbcinfo->mbctype[i + 1] & _M1))
 344     {
 345       if(mbcinfo->mbctype[i] & (C1_UPPER|C1_LOWER))
 346         mbcinfo->mbcasemap[i] = bufA[charcount];
 347       charcount++;
 348     }
 349   }
 350
 351   if (newcp == 932)   /* CP932 only - set _MP and _MS */
 352   {
 353     /* On Windows it's possible to calculate the _MP and _MS from CT_CTYPE1
 354      * and CT_CTYPE3. But as of Wine 0.9.43 we return wrong values what makes
 355      * it hard. As this is set only for codepage 932 we hardcode it what gives
 356      * also faster execution.
 357      */
 358     for (i = 161; i <= 165; i++)
 359       mbcinfo->mbctype[i + 1] |= _MP;
 360     for (i = 166; i <= 223; i++)
 361       mbcinfo->mbctype[i + 1] |= _MS;
 362   }
 363
 364   mbcinfo->mbcodepage = newcp;
 365   if(MSVCRT_locale && mbcinfo == MSVCRT_locale->mbcinfo)
 366     memcpy(MSVCRT_mbctype, MSVCRT_locale->mbcinfo->mbctype, sizeof(MSVCRT_mbctype));
 367
 368   return 0;
 369 }
 370
 371 /*********************************************************************
 372  *              _setmbcp (MSVCRT.@)
 373  */
 374 int CDECL _setmbcp(int cp)
 375 {
 376     return _setmbcp_l(cp, -1, NULL);
 377 }
 378
 379 /*********************************************************************
 380  *              _getmbcp (MSVCRT.@)
 381  */
 382 int CDECL _getmbcp(void)
 383 {
 384   return get_mbcinfo()->mbcodepage;
 385 }
 386
 387 /*********************************************************************
 388  *              _mbsnextc(MSVCRT.@)
 389  */
 390 unsigned int CDECL _mbsnextc(const unsigned char* str)
 391 {
 392   if(_ismbblead(*str))
 393     return *str << 8 | str[1];
 394   return *str;
 395 }
 396
 397 /*********************************************************************
 398  *              _mbctolower(MSVCRT.@)
 399  */
 400 unsigned int CDECL _mbctolower(unsigned int c)
 401 {
 402     if (_ismbblead(c))
 403     {
 404       FIXME("Handle MBC chars\n");
 405       return c;
 406     }
 407     return tolower(c); /* ASCII CP or SB char */
 408 }
 409
 410 /*********************************************************************
 411  *              _mbctoupper(MSVCRT.@)
 412  */
 413 unsigned int CDECL _mbctoupper(unsigned int c)
 414 {
 415     if (_ismbblead(c))
 416     {
 417       FIXME("Handle MBC chars\n");
 418       return c;
 419     }
 420     return toupper(c); /* ASCII CP or SB char */
 421 }
 422
 423 /*********************************************************************
 424  *              _mbctombb (MSVCRT.@)
 425  */
 426 unsigned int CDECL _mbctombb(unsigned int c)
 427 {
 428     unsigned int value;
 429
 430     if(get_mbcinfo()->mbcodepage == 932)
 431     {
 432         if(c >= 0x829f && c <= 0x82f1)    /* Hiragana */
 433             return mbctombb_932_kana[c - 0x829f];
 434         if(c >= 0x8340 && c <= 0x8396 && c != 0x837f)    /* Katakana */
 435             return mbctombb_932_kana[c - 0x8340 - (c >= 0x837f ? 1 : 0)];
 436         if(c >= 0x8140 && c <= 0x8197)    /* Punctuation */
 437         {
 438             value = mbctombb_932_punct[c - 0x8140];
 439             return value ? value : c;
 440         }
 441         if((c >= 0x824f && c <= 0x8258) || /* Fullwidth digits */
 442            (c >= 0x8260 && c <= 0x8279))   /* Fullwidth capitals letters */
 443             return c - 0x821f;
 444         if(c >= 0x8281 && c <= 0x829a)     /* Fullwidth small letters */
 445             return c - 0x8220;
 446         /* all other cases return c */
 447     }
 448     return c;
 449 }
 450
 451 /*********************************************************************
 452  *              _mbcjistojms(MSVCRT.@)
 453  *
 454  *              Converts a jis character to sjis.
 455  *              Based on description from
 456  *              http://www.slayers.ne.jp/~oouchi/code/jistosjis.html
 457  */
 458 unsigned int CDECL _mbcjistojms(unsigned int c)
 459 {
 460   /* Conversion takes place only when codepage is 932.
 461      In all other cases, c is returned unchanged */
 462   if(get_mbcinfo()->mbcodepage == 932)
 463   {
 464     if(HIBYTE(c) >= 0x21 && HIBYTE(c) <= 0x7e &&
 465        LOBYTE(c) >= 0x21 && LOBYTE(c) <= 0x7e)
 466     {
 467       if(HIBYTE(c) % 2)
 468         c += 0x1f;
 469       else
 470         c += 0x7d;
 471
 472       if(LOBYTE(c) >= 0x7F)
 473         c += 0x1;
 474
 475       c = (((HIBYTE(c) - 0x21)/2 + 0x81) << 8) | LOBYTE(c);
 476
 477       if(HIBYTE(c) > 0x9f)
 478         c += 0x4000;
 479     }
 480     else
 481       return 0; /* Codepage is 932, but c can't be converted */
 482   }
 483
 484   return c;
 485 }
 486
 487 /*********************************************************************
 488  *              _mbcjmstojis(MSVCRT.@)
 489  *
 490  *              Converts a sjis character to jis.
 491  */
 492 unsigned int CDECL _mbcjmstojis(unsigned int c)
 493 {
 494   /* Conversion takes place only when codepage is 932.
 495      In all other cases, c is returned unchanged */
 496   if(get_mbcinfo()->mbcodepage == 932)
 497   {
 498     if(_ismbclegal(c) && HIBYTE(c) < 0xf0)
 499     {
 500       if(HIBYTE(c) >= 0xe0)
 501         c -= 0x4000;
 502
 503       c = (((HIBYTE(c) - 0x81)*2 + 0x21) << 8) | LOBYTE(c);
 504
 505       if(LOBYTE(c) > 0x7f)
 506         c -= 0x1;
 507
 508       if(LOBYTE(c) > 0x9d)
 509         c += 0x83;
 510       else
 511         c -= 0x1f;
 512     }
 513     else
 514       return 0; /* Codepage is 932, but c can't be converted */
 515   }
 516
 517   return c;
 518 }
 519
 520 /*********************************************************************
 521  *              _mbsdec(MSVCRT.@)
 522  */
 523 unsigned char* CDECL _mbsdec(const unsigned char* start, const unsigned char* cur)
 524 {
 525   if(get_mbcinfo()->ismbcodepage)
 526     return (unsigned char *)(_ismbstrail(start,cur-1) ? cur - 2 : cur -1);
 527
 528   return (unsigned char *)cur - 1; /* ASCII CP or SB char */
 529 }
 530
 531 /*********************************************************************
 532  *              _mbclen(MSVCRT.@)
 533  */
 534 unsigned int CDECL _mbclen(const unsigned char* str)
 535 {
 536   return _ismbblead(*str) ? 2 : 1;
 537 }
 538
 539 /*********************************************************************
 540  *              _mbsinc(MSVCRT.@)
 541  */
 542 unsigned char* CDECL _mbsinc(const unsigned char* str)
 543 {
 544   return (unsigned char *)(str + _mbclen(str));
 545 }
 546
 547 /*********************************************************************
 548  *              _mbsninc(MSVCRT.@)
 549  */
 550 unsigned char* CDECL _mbsninc(const unsigned char* str, MSVCRT_size_t num)
 551 {
 552   if(!str)
 553     return NULL;
 554
 555   while (num > 0 && *str)
 556   {
 557     if (_ismbblead(*str))
 558     {
 559       if (!*(str+1))
 560          break;
 561       str++;
 562     }
 563     str++;
 564     num--;
 565   }
 566
 567   return (unsigned char*)str;
 568 }
 569
 570 /*********************************************************************
 571  *              _mbslen(MSVCRT.@)
 572  */
 573 MSVCRT_size_t CDECL _mbslen(const unsigned char* str)
 574 {
 575   MSVCRT_size_t len = 0;
 576   while(*str)
 577   {
 578     if (_ismbblead(*str))
 579     {
 580       str++;
 581       if (!*str)  /* count only full chars */
 582         break;
 583     }
 584     str++;
 585     len++;
 586   }
 587   return len;
 588 }
 589
 590 /*********************************************************************
 591  *              _mbccpy(MSVCRT.@)
 592  */
 593 void CDECL _mbccpy(unsigned char* dest, const unsigned char* src)
 594 {
 595   *dest = *src;
 596   if(_ismbblead(*src))
 597     *++dest = *++src; /* MB char */
 598 }
 599
 600 /*********************************************************************
 601  *              _mbsncpy(MSVCRT.@)
 602  * REMARKS
 603  *  The parameter n is the number or characters to copy, not the size of
 604  *  the buffer. Use _mbsnbcpy for a function analogical to strncpy
 605  */
 606 unsigned char* CDECL _mbsncpy(unsigned char* dst, const unsigned char* src, MSVCRT_size_t n)
 607 {
 608   unsigned char* ret = dst;
 609   if(!n)
 610     return dst;
 611   if (get_mbcinfo()->ismbcodepage)
 612   {
 613     while (*src && n)
 614     {
 615       n--;
 616       if (_ismbblead(*src))
 617       {
 618         if (!*(src+1))
 619         {
 620             *dst++ = 0;
 621             *dst++ = 0;
 622             break;
 623         }
 624
 625         *dst++ = *src++;
 626       }
 627
 628       *dst++ = *src++;
 629     }
 630   }
 631   else
 632   {
 633     while (n)
 634     {
 635         n--;
 636         if (!(*dst++ = *src++)) break;
 637     }
 638   }
 639   while (n--) *dst++ = 0;
 640   return ret;
 641 }
 642
 643 /*********************************************************************
 644  *              _mbsnbcpy_s(MSVCRT.@)
 645  * REMARKS
 646  * Unlike _mbsnbcpy this function does not pad the rest of the dest
 647  * string with 0
 648  */
 649 int CDECL _mbsnbcpy_s(unsigned char* dst, MSVCRT_size_t size, const unsigned char* src, MSVCRT_size_t n)
 650 {
 651     MSVCRT_size_t pos = 0;
 652
 653     if(!dst || size == 0)
 654         return MSVCRT_EINVAL;
 655     if(!src)
 656     {
 657         dst[0] = '\0';
 658         return MSVCRT_EINVAL;
 659     }
 660     if(!n)
 661         return 0;
 662
 663     if(get_mbcinfo()->ismbcodepage)
 664     {
 665         int is_lead = 0;
 666         while (*src && n)
 667         {
 668             if(pos == size)
 669             {
 670                 dst[0] = '\0';
 671                 return MSVCRT_ERANGE;
 672             }
 673             is_lead = (!is_lead && _ismbblead(*src));
 674             n--;
 675             dst[pos++] = *src++;
 676         }
 677
 678         if (is_lead) /* if string ends with a lead, remove it */
 679             dst[pos - 1] = 0;
 680     }
 681     else
 682     {
 683         while (n)
 684         {
 685             n--;
 686             if(pos == size)
 687             {
 688                 dst[0] = '\0';
 689                 return MSVCRT_ERANGE;
 690             }
 691
 692             if(!(*src)) break;
 693             dst[pos++] = *src++;
 694         }
 695     }
 696
 697     if(pos < size)
 698         dst[pos] = '\0';
 699     else
 700     {
 701         dst[0] = '\0';
 702         return MSVCRT_ERANGE;
 703     }
 704
 705     return 0;
 706 }
 707
 708 /*********************************************************************
 709  *              _mbsnbcpy(MSVCRT.@)
 710  * REMARKS
 711  *  Like strncpy this function doesn't enforce the string to be
 712  *  NUL-terminated
 713  */
 714 unsigned char* CDECL _mbsnbcpy(unsigned char* dst, const unsigned char* src, MSVCRT_size_t n)
 715 {
 716   unsigned char* ret = dst;
 717   if(!n)
 718     return dst;
 719   if(get_mbcinfo()->ismbcodepage)
 720   {
 721     int is_lead = 0;
 722     while (*src && n)
 723     {
 724       is_lead = (!is_lead && _ismbblead(*src));
 725       n--;
 726       *dst++ = *src++;
 727     }
 728
 729     if (is_lead) /* if string ends with a lead, remove it */
 730         *(dst - 1) = 0;
 731   }
 732   else
 733   {
 734     while (n)
 735     {
 736         n--;
 737         if (!(*dst++ = *src++)) break;
 738     }
 739   }
 740   while (n--) *dst++ = 0;
 741   return ret;
 742 }
 743
 744 /*********************************************************************
 745  *              _mbscmp(MSVCRT.@)
 746  */
 747 int CDECL _mbscmp(const unsigned char* str, const unsigned char* cmp)
 748 {
 749   if(get_mbcinfo()->ismbcodepage)
 750   {
 751     unsigned int strc, cmpc;
 752     do {
 753       if(!*str)
 754         return *cmp ? -1 : 0;
 755       if(!*cmp)
 756         return 1;
 757       strc = _mbsnextc(str);
 758       cmpc = _mbsnextc(cmp);
 759       if(strc != cmpc)
 760         return strc < cmpc ? -1 : 1;
 761       str +=(strc > 255) ? 2 : 1;
 762       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 763     } while(1);
 764   }
 765   return u_strcmp(str, cmp); /* ASCII CP */
 766 }
 767
 768 /*********************************************************************
 769  *              _mbsicoll(MSVCRT.@)
 770  * FIXME: handle locales.
 771  */
 772 int CDECL _mbsicoll(const unsigned char* str, const unsigned char* cmp)
 773 {
 774   if(get_mbcinfo()->ismbcodepage)
 775   {
 776     unsigned int strc, cmpc;
 777     do {
 778       if(!*str)
 779         return *cmp ? -1 : 0;
 780       if(!*cmp)
 781         return 1;
 782       strc = _mbctolower(_mbsnextc(str));
 783       cmpc = _mbctolower(_mbsnextc(cmp));
 784       if(strc != cmpc)
 785         return strc < cmpc ? -1 : 1;
 786       str +=(strc > 255) ? 2 : 1;
 787       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 788     } while(1);
 789   }
 790   return u_strcasecmp(str, cmp); /* ASCII CP */
 791 }
 792
 793 /*********************************************************************
 794  *              _mbscoll(MSVCRT.@)
 795  * Performs a case-sensitive comparison according to the current code page
 796  * RETURN
 797  *   _NLSCMPERROR if error
 798  * FIXME: handle locales.
 799  */
 800 int CDECL _mbscoll(const unsigned char* str, const unsigned char* cmp)
 801 {
 802   if(get_mbcinfo()->ismbcodepage)
 803   {
 804     unsigned int strc, cmpc;
 805     do {
 806       if(!*str)
 807         return *cmp ? -1 : 0;
 808       if(!*cmp)
 809         return 1;
 810       strc = _mbsnextc(str);
 811       cmpc = _mbsnextc(cmp);
 812       if(strc != cmpc)
 813         return strc < cmpc ? -1 : 1;
 814       str +=(strc > 255) ? 2 : 1;
 815       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 816     } while(1);
 817   }
 818   return u_strcmp(str, cmp); /* ASCII CP */
 819 }
 820
 821
 822 /*********************************************************************
 823  *              _mbsicmp(MSVCRT.@)
 824  */
 825 int CDECL _mbsicmp(const unsigned char* str, const unsigned char* cmp)
 826 {
 827   if(get_mbcinfo()->ismbcodepage)
 828   {
 829     unsigned int strc, cmpc;
 830     do {
 831       if(!*str)
 832         return *cmp ? -1 : 0;
 833       if(!*cmp)
 834         return 1;
 835       strc = _mbctolower(_mbsnextc(str));
 836       cmpc = _mbctolower(_mbsnextc(cmp));
 837       if(strc != cmpc)
 838         return strc < cmpc ? -1 : 1;
 839       str +=(strc > 255) ? 2 : 1;
 840       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 841     } while(1);
 842   }
 843   return u_strcasecmp(str, cmp); /* ASCII CP */
 844 }
 845
 846 /*********************************************************************
 847  *              _mbsncmp(MSVCRT.@)
 848  */
 849 int CDECL _mbsncmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 850 {
 851   if(!len)
 852     return 0;
 853
 854   if(get_mbcinfo()->ismbcodepage)
 855   {
 856     unsigned int strc, cmpc;
 857     while(len--)
 858     {
 859       int inc;
 860       if(!*str)
 861         return *cmp ? -1 : 0;
 862       if(!*cmp)
 863         return 1;
 864       strc = _mbsnextc(str);
 865       cmpc = _mbsnextc(cmp);
 866       if(strc != cmpc)
 867         return strc < cmpc ? -1 : 1;
 868       inc=(strc > 255) ? 2 : 1; /* Equal, use same increment */
 869       str += inc;
 870       cmp += inc;
 871     }
 872     return 0; /* Matched len chars */
 873   }
 874   return u_strncmp(str, cmp, len); /* ASCII CP */
 875 }
 876
 877 /*********************************************************************
 878  *              _mbsnbcmp(MSVCRT.@)
 879  */
 880 int CDECL _mbsnbcmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 881 {
 882   if (!len)
 883     return 0;
 884   if(get_mbcinfo()->ismbcodepage)
 885   {
 886     unsigned int strc, cmpc;
 887     while (len)
 888     {
 889       int clen;
 890       if(!*str)
 891         return *cmp ? -1 : 0;
 892       if(!*cmp)
 893         return 1;
 894       if (_ismbblead(*str))
 895       {
 896         strc=(len>=2)?_mbsnextc(str):0;
 897         clen=2;
 898       }
 899       else
 900       {
 901         strc=*str;
 902         clen=1;
 903       }
 904       if (_ismbblead(*cmp))
 905         cmpc=(len>=2)?_mbsnextc(cmp):0;
 906       else
 907         cmpc=*cmp;
 908       if(strc != cmpc)
 909         return strc < cmpc ? -1 : 1;
 910       len -= clen;
 911       str += clen;
 912       cmp += clen;
 913     }
 914     return 0; /* Matched len chars */
 915   }
 916   return u_strncmp(str,cmp,len);
 917 }
 918
 919 /*********************************************************************
 920  *              _mbsnicmp(MSVCRT.@)
 921  *
 922  * Compare two multibyte strings case insensitively to 'len' characters.
 923  */
 924 int CDECL _mbsnicmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 925 {
 926   /* FIXME: No tolower() for mb strings yet */
 927   if(get_mbcinfo()->ismbcodepage)
 928   {
 929     unsigned int strc, cmpc;
 930     while(len--)
 931     {
 932       if(!*str)
 933         return *cmp ? -1 : 0;
 934       if(!*cmp)
 935         return 1;
 936       strc = _mbctolower(_mbsnextc(str));
 937       cmpc = _mbctolower(_mbsnextc(cmp));
 938       if(strc != cmpc)
 939         return strc < cmpc ? -1 : 1;
 940       str +=(strc > 255) ? 2 : 1;
 941       cmp +=(strc > 255) ? 2 : 1; /* Equal, use same increment */
 942     }
 943     return 0; /* Matched len chars */
 944   }
 945   return u_strncasecmp(str, cmp, len); /* ASCII CP */
 946 }
 947
 948 /*********************************************************************
 949  *              _mbsnbicmp(MSVCRT.@)
 950  */
 951 int CDECL _mbsnbicmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 952 {
 953   if (!len)
 954     return 0;
 955   if(get_mbcinfo()->ismbcodepage)
 956   {
 957     unsigned int strc, cmpc;
 958     while (len)
 959     {
 960       int clen;
 961       if(!*str)
 962         return *cmp ? -1 : 0;
 963       if(!*cmp)
 964         return 1;
 965       if (_ismbblead(*str))
 966       {
 967         strc=(len>=2)?_mbsnextc(str):0;
 968         clen=2;
 969       }
 970       else
 971       {
 972         strc=*str;
 973         clen=1;
 974       }
 975       if (_ismbblead(*cmp))
 976         cmpc=(len>=2)?_mbsnextc(cmp):0;
 977       else
 978         cmpc=*cmp;
 979       strc = _mbctolower(strc);
 980       cmpc = _mbctolower(cmpc);
 981       if(strc != cmpc)
 982         return strc < cmpc ? -1 : 1;
 983       len -= clen;
 984       str += clen;
 985       cmp += clen;
 986     }
 987     return 0; /* Matched len bytes */
 988   }
 989   return u_strncasecmp(str,cmp,len);
 990 }
 991
 992 /*********************************************************************
 993  *              _mbscat (MSVCRT.@)
 994  */
 995 unsigned char * CDECL _mbscat( unsigned char *dst, const unsigned char *src )
 996 {
 997     strcat( (char *)dst, (const char *)src );
 998     return dst;
 999 }
1000
1001 /*********************************************************************
1002  *              _mbscpy (MSVCRT.@)
1003  */
1004 unsigned char* CDECL _mbscpy( unsigned char *dst, const unsigned char *src )
1005 {
1006     strcpy( (char *)dst, (const char *)src );
1007     return dst;
1008 }
1009
1010 /*********************************************************************
1011  *              _mbsstr (MSVCRT.@)
1012  */
1013 unsigned char * CDECL _mbsstr(const unsigned char *haystack, const unsigned char *needle)
1014 {
1015     return (unsigned char *)strstr( (const char *)haystack, (const char *)needle );
1016 }
1017
1018 /*********************************************************************
1019  *              _mbschr(MSVCRT.@)
1020  *
1021  * Find a multibyte character in a multibyte string.
1022  */
1023 unsigned char* CDECL _mbschr(const unsigned char* s, unsigned int x)
1024 {
1025   if(get_mbcinfo()->ismbcodepage)
1026   {
1027     unsigned int c;
1028     while (1)
1029     {
1030       c = _mbsnextc(s);
1031       if (c == x)
1032         return (unsigned char*)s;
1033       if (!c)
1034         return NULL;
1035       s += c > 255 ? 2 : 1;
1036     }
1037   }
1038   return u_strchr(s, x); /* ASCII CP */
1039 }
1040
1041 /*********************************************************************
1042  *              _mbsrchr(MSVCRT.@)
1043  */
1044 unsigned char* CDECL _mbsrchr(const unsigned char* s, unsigned int x)
1045 {
1046   if(get_mbcinfo()->ismbcodepage)
1047   {
1048     unsigned int c;
1049     unsigned char* match=NULL;
1050     if(!s)
1051       return NULL;
1052     while (1) {
1053       c = _mbsnextc(s);
1054       if (c == x)
1055         match=(unsigned char*)s;
1056       if (!c)
1057         return match;
1058       s +=(c > 255) ? 2 : 1;
1059     }
1060   }
1061   return u_strrchr(s, x);
1062 }
1063
1064 /*********************************************************************
1065  *              _mbstok(MSVCRT.@)
1066  *
1067  * Find and extract tokens from strings
1068  */
1069 unsigned char* CDECL _mbstok(unsigned char *str, const unsigned char *delim)
1070 {
1071     thread_data_t *data = msvcrt_get_thread_data();
1072     unsigned char *ret;
1073
1074     if(get_mbcinfo()->ismbcodepage)
1075     {
1076         unsigned int c;
1077
1078         if (!str)
1079             if (!(str = data->mbstok_next)) return NULL;
1080
1081         while ((c = _mbsnextc(str)) && _mbschr(delim, c)) {
1082             str += c > 255 ? 2 : 1;
1083         }
1084         if (!*str) return NULL;
1085         ret = str++;
1086         while ((c = _mbsnextc(str)) && !_mbschr(delim, c)) {
1087             str += c > 255 ? 2 : 1;
1088         }
1089         if (*str) {
1090             *str++ = 0;
1091             if (c > 255) *str++ = 0;
1092         }
1093         data->mbstok_next = str;
1094         return ret;
1095     }
1096     return u_strtok(str, delim); /* ASCII CP */
1097 }
1098
1099 /*********************************************************************
1100  *              _mbbtombc(MSVCRT.@)
1101  */
1102 unsigned int CDECL _mbbtombc(unsigned int c)
1103 {
1104   if(get_mbcinfo()->mbcodepage == 932)
1105   {
1106     if(c >= 0x20 && c <= 0x7e) {
1107       if((c >= 0x41 && c <= 0x5a) || (c >= 0x61 && c <= 0x7a) || (c >= 0x30 && c <= 0x39))
1108         return mbbtombc_932[c - 0x20] | 0x8200;
1109       else
1110         return mbbtombc_932[c - 0x20] | 0x8100;
1111     }
1112     else if(c >= 0xa1 && c <= 0xdf) {
1113       if(c >= 0xa6 && c <= 0xdd && c != 0xb0)
1114         return mbbtombc_932[c - 0xa1 + 0x5f] | 0x8300;
1115       else
1116         return mbbtombc_932[c - 0xa1 + 0x5f] | 0x8100;
1117     }
1118   }
1119   return c;  /* not Japanese or no MB char */
1120 }
1121
1122 /*********************************************************************
1123  *              _mbbtype(MSVCRT.@)
1124  */
1125 int CDECL _mbbtype(unsigned char c, int type)
1126 {
1127     if (type == 1)
1128     {
1129         if ((c >= 0x20 && c <= 0x7e) || (c >= 0xa1 && c <= 0xdf))
1130             return _MBC_SINGLE;
1131         else if ((c >= 0x40 && c <= 0x7e) || (c >= 0x80 && c <= 0xfc))
1132             return _MBC_TRAIL;
1133         else
1134             return _MBC_ILLEGAL;
1135     }
1136     else
1137     {
1138         if ((c >= 0x20 && c <= 0x7e) || (c >= 0xa1 && c <= 0xdf))
1139             return _MBC_SINGLE;
1140         else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc))
1141             return _MBC_LEAD;
1142         else
1143             return _MBC_ILLEGAL;
1144     }
1145 }
1146
1147 /*********************************************************************
1148  *              _ismbbkana(MSVCRT.@)
1149  */
1150 int CDECL _ismbbkana(unsigned int c)
1151 {
1152   /* FIXME: use lc_ctype when supported, not lc_all */
1153   if(get_mbcinfo()->mbcodepage == 932)
1154   {
1155     /* Japanese/Katakana, CP 932 */
1156     return (c >= 0xa1 && c <= 0xdf);
1157   }
1158   return 0;
1159 }
1160
1161 /*********************************************************************
1162  *              _ismbcdigit(MSVCRT.@)
1163  */
1164 int CDECL _ismbcdigit(unsigned int ch)
1165 {
1166     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1167     return (get_char_typeW( wch ) & C1_DIGIT);
1168 }
1169
1170 /*********************************************************************
1171  *              _ismbcgraph(MSVCRT.@)
1172  */
1173 int CDECL _ismbcgraph(unsigned int ch)
1174 {
1175     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1176     return (get_char_typeW( wch ) & (C1_UPPER | C1_LOWER | C1_DIGIT | C1_PUNCT | C1_ALPHA));
1177 }
1178
1179 /*********************************************************************
1180  *              _ismbcalpha (MSVCRT.@)
1181  */
1182 int CDECL _ismbcalpha(unsigned int ch)
1183 {
1184     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1185     return (get_char_typeW( wch ) & C1_ALPHA);
1186 }
1187
1188 /*********************************************************************
1189  *              _ismbclower (MSVCRT.@)
1190  */
1191 int CDECL _ismbclower(unsigned int ch)
1192 {
1193     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1194     return (get_char_typeW( wch ) & C1_UPPER);
1195 }
1196
1197 /*********************************************************************
1198  *              _ismbcupper (MSVCRT.@)
1199  */
1200 int CDECL _ismbcupper(unsigned int ch)
1201 {
1202     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1203     return (get_char_typeW( wch ) & C1_LOWER);
1204 }
1205
1206 /*********************************************************************
1207  *              _ismbcsymbol(MSVCRT.@)
1208  */
1209 int CDECL _ismbcsymbol(unsigned int ch)
1210 {
1211     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1212     WORD ctype;
1213     if (!GetStringTypeW(CT_CTYPE3, &wch, 1, &ctype))
1214     {
1215         WARN("GetStringTypeW failed on %x\n", ch);
1216         return 0;
1217     }
1218     return ((ctype & C3_SYMBOL) != 0);
1219 }
1220
1221 /*********************************************************************
1222  *              _ismbcalnum (MSVCRT.@)
1223  */
1224 int CDECL _ismbcalnum(unsigned int ch)
1225 {
1226     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1227     return (get_char_typeW( wch ) & (C1_ALPHA | C1_DIGIT));
1228 }
1229
1230 /*********************************************************************
1231  *              _ismbcspace (MSVCRT.@)
1232  */
1233 int CDECL _ismbcspace(unsigned int ch)
1234 {
1235     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1236     return (get_char_typeW( wch ) & C1_SPACE);
1237 }
1238
1239 /*********************************************************************
1240  *              _ismbcprint (MSVCRT.@)
1241  */
1242 int CDECL _ismbcprint(unsigned int ch)
1243 {
1244     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1245     return (get_char_typeW( wch ) & (C1_UPPER | C1_LOWER | C1_DIGIT | C1_PUNCT | C1_ALPHA | C1_SPACE));
1246 }
1247
1248 /*********************************************************************
1249  *              _ismbcpunct(MSVCRT.@)
1250  */
1251 int CDECL _ismbcpunct(unsigned int ch)
1252 {
1253     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1254     return (get_char_typeW( wch ) & C1_PUNCT);
1255 }
1256
1257 /*********************************************************************
1258  *              _ismbchira(MSVCRT.@)
1259  */
1260 int CDECL _ismbchira(unsigned int c)
1261 {
1262   /* FIXME: use lc_ctype when supported, not lc_all */
1263   if(get_mbcinfo()->mbcodepage == 932)
1264   {
1265     /* Japanese/Hiragana, CP 932 */
1266     return (c >= 0x829f && c <= 0x82f1);
1267   }
1268   return 0;
1269 }
1270
1271 /*********************************************************************
1272  *              _ismbckata(MSVCRT.@)
1273  */
1274 int CDECL _ismbckata(unsigned int c)
1275 {
1276   /* FIXME: use lc_ctype when supported, not lc_all */
1277   if(get_mbcinfo()->mbcodepage == 932)
1278   {
1279     if(c < 256)
1280       return _ismbbkana(c);
1281     /* Japanese/Katakana, CP 932 */
1282     return (c >= 0x8340 && c <= 0x8396 && c != 0x837f);
1283   }
1284   return 0;
1285 }
1286
1287 /*********************************************************************
1288  *              _ismbblead(MSVCRT.@)
1289  */
1290 int CDECL _ismbblead(unsigned int c)
1291 {
1292   return (get_mbcinfo()->mbctype[(c&0xff) + 1] & _M1) != 0;
1293 }
1294
1295
1296 /*********************************************************************
1297  *              _ismbbtrail(MSVCRT.@)
1298  */
1299 int CDECL _ismbbtrail(unsigned int c)
1300 {
1301   return (get_mbcinfo()->mbctype[(c&0xff) + 1] & _M2) != 0;
1302 }
1303
1304 /*********************************************************************
1305  *              _ismbclegal(MSVCRT.@)
1306  */
1307 int CDECL _ismbclegal(unsigned int c)
1308 {
1309     return _ismbblead(HIBYTE(c)) && _ismbbtrail(LOBYTE(c));
1310 }
1311
1312 /*********************************************************************
1313  *              _ismbslead(MSVCRT.@)
1314  */
1315 int CDECL _ismbslead(const unsigned char* start, const unsigned char* str)
1316 {
1317   int lead = 0;
1318
1319   if(!get_mbcinfo()->ismbcodepage)
1320     return 0;
1321
1322   /* Lead bytes can also be trail bytes so we need to analyse the string
1323    */
1324   while (start <= str)
1325   {
1326     if (!*start)
1327       return 0;
1328     lead = !lead && _ismbblead(*start);
1329     start++;
1330   }
1331
1332   return lead ? -1 : 0;
1333 }
1334
1335 /*********************************************************************
1336  *              _ismbstrail(MSVCRT.@)
1337  */
1338 int CDECL _ismbstrail(const unsigned char* start, const unsigned char* str)
1339 {
1340   /* Note: this function doesn't check _ismbbtrail */
1341   if ((str > start) && _ismbslead(start, str-1))
1342     return -1;
1343   else
1344     return 0;
1345 }
1346
1347 /*********************************************************************
1348  *              _mbsbtype (MSVCRT.@)
1349  */
1350 int CDECL _mbsbtype(const unsigned char *str, MSVCRT_size_t count)
1351 {
1352   int lead = 0;
1353   const unsigned char *end = str + count;
1354
1355   /* Lead bytes can also be trail bytes so we need to analyse the string.
1356    * Also we must return _MBC_ILLEGAL for chars past the end of the string
1357    */
1358   while (str < end) /* Note: we skip the last byte - will check after the loop */
1359   {
1360     if (!*str)
1361       return _MBC_ILLEGAL;
1362     lead = get_mbcinfo()->ismbcodepage && !lead && _ismbblead(*str);
1363     str++;
1364   }
1365
1366   if (lead)
1367     if (_ismbbtrail(*str))
1368       return _MBC_TRAIL;
1369     else
1370       return _MBC_ILLEGAL;
1371   else
1372     if (_ismbblead(*str))
1373       return _MBC_LEAD;
1374     else
1375       return _MBC_SINGLE;
1376 }
1377
1378 /*********************************************************************
1379  *              _mbsset(MSVCRT.@)
1380  */
1381 unsigned char* CDECL _mbsset(unsigned char* str, unsigned int c)
1382 {
1383   unsigned char* ret = str;
1384
1385   if(!get_mbcinfo()->ismbcodepage || c < 256)
1386     return u__strset(str, c); /* ASCII CP or SB char */
1387
1388   c &= 0xffff; /* Strip high bits */
1389
1390   while(str[0] && str[1])
1391   {
1392     *str++ = c >> 8;
1393     *str++ = c & 0xff;
1394   }
1395   if(str[0])
1396     str[0] = '\0'; /* FIXME: OK to shorten? */
1397
1398   return ret;
1399 }
1400
1401 /*********************************************************************
1402  *              _mbsnbset(MSVCRT.@)
1403  */
1404 unsigned char* CDECL _mbsnbset(unsigned char *str, unsigned int c, MSVCRT_size_t len)
1405 {
1406     unsigned char *ret = str;
1407
1408     if(!len)
1409         return ret;
1410
1411     if(!get_mbcinfo()->ismbcodepage || c < 256)
1412         return u__strnset(str, c, len); /* ASCII CP or SB char */
1413
1414     c &= 0xffff; /* Strip high bits */
1415
1416     while(str[0] && str[1] && (len > 1))
1417     {
1418         *str++ = c >> 8;
1419         len--;
1420         *str++ = c & 0xff;
1421         len--;
1422     }
1423     if(len && str[0]) {
1424         /* as per msdn pad with a blank character */
1425         str[0] = ' ';
1426     }
1427
1428     return ret;
1429 }
1430
1431 /*********************************************************************
1432  *              _mbsnset(MSVCRT.@)
1433  */
1434 unsigned char* CDECL _mbsnset(unsigned char* str, unsigned int c, MSVCRT_size_t len)
1435 {
1436   unsigned char *ret = str;
1437
1438   if(!len)
1439     return ret;
1440
1441   if(!get_mbcinfo()->ismbcodepage || c < 256)
1442     return u__strnset(str, c, len); /* ASCII CP or SB char */
1443
1444   c &= 0xffff; /* Strip high bits */
1445
1446   while(str[0] && str[1] && len--)
1447   {
1448     *str++ = c >> 8;
1449     *str++ = c & 0xff;
1450   }
1451   if(len && str[0])
1452     str[0] = '\0'; /* FIXME: OK to shorten? */
1453
1454   return ret;
1455 }
1456
1457 /*********************************************************************
1458  *              _mbsnccnt(MSVCRT.@)
1459  * 'c' is for 'character'.
1460  */
1461 MSVCRT_size_t CDECL _mbsnccnt(const unsigned char* str, MSVCRT_size_t len)
1462 {
1463   MSVCRT_size_t ret;
1464   if(get_mbcinfo()->ismbcodepage)
1465   {
1466     ret=0;
1467     while(*str && len-- > 0)
1468     {
1469       if(_ismbblead(*str))
1470       {
1471         if (!len)
1472           break;
1473         len--;
1474         str++;
1475       }
1476       str++;
1477       ret++;
1478     }
1479     return ret;
1480   }
1481   ret=u_strlen(str);
1482   return min(ret, len); /* ASCII CP */
1483 }
1484
1485 /*********************************************************************
1486  *              _mbsnbcnt(MSVCRT.@)
1487  * 'b' is for byte count.
1488  */
1489 MSVCRT_size_t CDECL _mbsnbcnt(const unsigned char* str, MSVCRT_size_t len)
1490 {
1491   MSVCRT_size_t ret;
1492   if(get_mbcinfo()->ismbcodepage)
1493   {
1494     const unsigned char* xstr = str;
1495     while(*xstr && len-- > 0)
1496     {
1497       if (_ismbblead(*xstr++))
1498         xstr++;
1499     }
1500     return xstr-str;
1501   }
1502   ret=u_strlen(str);
1503   return min(ret, len); /* ASCII CP */
1504 }
1505
1506 /*********************************************************************
1507  *              _mbsnbcat(MSVCRT.@)
1508  */
1509 unsigned char* CDECL _mbsnbcat(unsigned char* dst, const unsigned char* src, MSVCRT_size_t len)
1510 {
1511     if(get_mbcinfo()->ismbcodepage)
1512     {
1513         unsigned char *res = dst;
1514         while (*dst) {
1515             if (_ismbblead(*dst++)) {
1516                 if (*dst) {
1517                     dst++;
1518                 } else {
1519                     /* as per msdn overwrite the lead byte in front of '\0' */
1520                     dst--;
1521                     break;
1522                 }
1523             }
1524         }
1525         while (*src && len--) *dst++ = *src++;
1526         *dst = '\0';
1527         return res;
1528     }
1529     return u_strncat(dst, src, len); /* ASCII CP */
1530 }
1531
1532 int CDECL _mbsnbcat_s(unsigned char *dst, MSVCRT_size_t size, const unsigned char *src, MSVCRT_size_t len)
1533 {
1534     unsigned char *ptr = dst;
1535     MSVCRT_size_t i;
1536
1537     if (!dst && !size && !src && !len)
1538         return 0;
1539
1540     if (!dst || !size || !src)
1541     {
1542         if (dst && size)
1543             *dst = '\0';
1544
1545         *MSVCRT__errno() = MSVCRT_EINVAL;
1546         return MSVCRT_EINVAL;
1547     }
1548
1549     /* Find the null terminator of the destination buffer. */
1550     while (size && *ptr)
1551         size--, ptr++;
1552
1553     if (!size)
1554     {
1555         *dst = '\0';
1556         *MSVCRT__errno() = MSVCRT_EINVAL;
1557         return MSVCRT_EINVAL;
1558     }
1559
1560     /* If necessary, check that the character preceding the null terminator is
1561      * a lead byte and move the pointer back by one for later overwrite. */
1562     if (ptr != dst && get_mbcinfo()->ismbcodepage && _ismbblead(*(ptr - 1)))
1563         size++, ptr--;
1564
1565     for (i = 0; *src && i < len; i++)
1566     {
1567         *ptr++ = *src++;
1568         size--;
1569
1570         if (!size)
1571         {
1572             *dst = '\0';
1573             *MSVCRT__errno() = MSVCRT_ERANGE;
1574             return MSVCRT_ERANGE;
1575         }
1576     }
1577
1578     *ptr = '\0';
1579     return 0;
1580 }
1581
1582 /*********************************************************************
1583  *              _mbsncat(MSVCRT.@)
1584  */
1585 unsigned char* CDECL _mbsncat(unsigned char* dst, const unsigned char* src, MSVCRT_size_t len)
1586 {
1587   if(get_mbcinfo()->ismbcodepage)
1588   {
1589     unsigned char *res = dst;
1590     while (*dst)
1591     {
1592       if (_ismbblead(*dst++))
1593         dst++;
1594     }
1595     while (*src && len--)
1596     {
1597       *dst++ = *src;
1598       if(_ismbblead(*src++))
1599         *dst++ = *src++;
1600     }
1601     *dst = '\0';
1602     return res;
1603   }
1604   return u_strncat(dst, src, len); /* ASCII CP */
1605 }
1606
1607
1608 /*********************************************************************
1609  *              _mbslwr(MSVCRT.@)
1610  */
1611 unsigned char* CDECL _mbslwr(unsigned char* s)
1612 {
1613   unsigned char *ret = s;
1614   if (!s)
1615     return NULL;
1616   if (get_mbcinfo()->ismbcodepage)
1617   {
1618     unsigned int c;
1619     while (*s)
1620     {
1621       c = _mbctolower(_mbsnextc(s));
1622       /* Note that I assume that the size of the character is unchanged */
1623       if (c > 255)
1624       {
1625           *s++=(c>>8);
1626           c=c & 0xff;
1627       }
1628       *s++=c;
1629     }
1630   }
1631   else for ( ; *s; s++) *s = tolower(*s);
1632   return ret;
1633 }
1634
1635
1636 /*********************************************************************
1637  *              _mbslwr_s(MSVCRT.@)
1638  */
1639 int CDECL _mbslwr_s(unsigned char* s, MSVCRT_size_t len)
1640 {
1641   if (!s && !len)
1642   {
1643     return 0;
1644   }
1645   else if (!s || !len)
1646   {
1647     *MSVCRT__errno() = MSVCRT_EINVAL;
1648     return MSVCRT_EINVAL;
1649   }
1650   if (get_mbcinfo()->ismbcodepage)
1651   {
1652     unsigned int c;
1653     for ( ; *s && len > 0; len--)
1654     {
1655       c = _mbctolower(_mbsnextc(s));
1656       /* Note that I assume that the size of the character is unchanged */
1657       if (c > 255)
1658       {
1659           *s++=(c>>8);
1660           c=c & 0xff;
1661       }
1662       *s++=c;
1663     }
1664   }
1665   else for ( ; *s && len > 0; s++, len--) *s = tolower(*s);
1666   if (*s)
1667   {
1668     *s = '\0';
1669     *MSVCRT__errno() = MSVCRT_EINVAL;
1670     return MSVCRT_EINVAL;
1671   }
1672   return 0;
1673 }
1674
1675
1676 /*********************************************************************
1677  *              _mbsupr(MSVCRT.@)
1678  */
1679 unsigned char* CDECL _mbsupr(unsigned char* s)
1680 {
1681   unsigned char *ret = s;
1682   if (!s)
1683     return NULL;
1684   if (get_mbcinfo()->ismbcodepage)
1685   {
1686     unsigned int c;
1687     while (*s)
1688     {
1689       c = _mbctoupper(_mbsnextc(s));
1690       /* Note that I assume that the size of the character is unchanged */
1691       if (c > 255)
1692       {
1693           *s++=(c>>8);
1694           c=c & 0xff;
1695       }
1696       *s++=c;
1697     }
1698   }
1699   else for ( ; *s; s++) *s = toupper(*s);
1700   return ret;
1701 }
1702
1703
1704 /*********************************************************************
1705  *              _mbsupr_s(MSVCRT.@)
1706  */
1707 int CDECL _mbsupr_s(unsigned char* s, MSVCRT_size_t len)
1708 {
1709   if (!s && !len)
1710   {
1711     return 0;
1712   }
1713   else if (!s || !len)
1714   {
1715     *MSVCRT__errno() = MSVCRT_EINVAL;
1716     return MSVCRT_EINVAL;
1717   }
1718   if (get_mbcinfo()->ismbcodepage)
1719   {
1720     unsigned int c;
1721     for ( ; *s && len > 0; len--)
1722     {
1723       c = _mbctoupper(_mbsnextc(s));
1724       /* Note that I assume that the size of the character is unchanged */
1725       if (c > 255)
1726       {
1727           *s++=(c>>8);
1728           c=c & 0xff;
1729       }
1730       *s++=c;
1731     }
1732   }
1733   else for ( ; *s && len > 0; s++, len--) *s = toupper(*s);
1734   if (*s)
1735   {
1736     *s = '\0';
1737     *MSVCRT__errno() = MSVCRT_EINVAL;
1738     return MSVCRT_EINVAL;
1739   }
1740   return 0;
1741 }
1742
1743
1744 /*********************************************************************
1745  *              _mbsspn (MSVCRT.@)
1746  */
1747 MSVCRT_size_t CDECL _mbsspn(const unsigned char* string, const unsigned char* set)
1748 {
1749     const unsigned char *p, *q;
1750
1751     for (p = string; *p; p++)
1752     {
1753         if (_ismbblead(*p))
1754         {
1755             for (q = set; *q; q++)
1756             {
1757                 if (!q[1])
1758                     break;
1759                 if ((*p == *q) &&  (p[1] == q[1]))
1760                     break;
1761                 q++;
1762             }
1763             if (!q[0] || !q[1]) break;
1764         }
1765         else
1766         {
1767             for (q = set; *q; q++)
1768                 if (*p == *q)
1769                     break;
1770             if (!*q) break;
1771         }
1772     }
1773     return p - string;
1774 }
1775
1776 /*********************************************************************
1777  *              _mbsspnp (MSVCRT.@)
1778  */
1779 unsigned char* CDECL _mbsspnp(const unsigned char* string, const unsigned char* set)
1780 {
1781     const unsigned char *p, *q;
1782
1783     for (p = string; *p; p++)
1784     {
1785         if (_ismbblead(*p))
1786         {
1787             for (q = set; *q; q++)
1788             {
1789                 if (!q[1])
1790                     break;
1791                 if ((*p == *q) &&  (p[1] == q[1]))
1792                     break;
1793                 q++;
1794             }
1795             if (!q[0] || !q[1]) break;
1796         }
1797         else
1798         {
1799             for (q = set; *q; q++)
1800                 if (*p == *q)
1801                     break;
1802             if (!*q) break;
1803         }
1804     }
1805     if (*p == '\0')
1806         return NULL;
1807     return (unsigned char *)p;
1808 }
1809
1810 /*********************************************************************
1811  *              _mbscspn(MSVCRT.@)
1812  */
1813 MSVCRT_size_t CDECL _mbscspn(const unsigned char* str, const unsigned char* cmp)
1814 {
1815   if (get_mbcinfo()->ismbcodepage)
1816     FIXME("don't handle double character case\n");
1817   return u_strcspn(str, cmp);
1818 }
1819
1820 /*********************************************************************
1821  *              _mbsrev (MSVCRT.@)
1822  */
1823 unsigned char* CDECL _mbsrev(unsigned char* str)
1824 {
1825     int i, len = _mbslen(str);
1826     unsigned char *p, *temp=MSVCRT_malloc(len*2);
1827
1828     if(!temp)
1829         return str;
1830
1831     /* unpack multibyte string to temp buffer */
1832     p=str;
1833     for(i=0; i<len; i++)
1834     {
1835         if (_ismbblead(*p))
1836         {
1837             temp[i*2]=*p++;
1838             temp[i*2+1]=*p++;
1839         }
1840         else
1841         {
1842             temp[i*2]=*p++;
1843             temp[i*2+1]=0;
1844         }
1845     }
1846
1847     /* repack it in the reverse order */
1848     p=str;
1849     for(i=len-1; i>=0; i--)
1850     {
1851         if(_ismbblead(temp[i*2]))
1852         {
1853             *p++=temp[i*2];
1854             *p++=temp[i*2+1];
1855         }
1856         else
1857         {
1858             *p++=temp[i*2];
1859         }
1860     }
1861
1862     MSVCRT_free(temp);
1863
1864     return str;
1865 }
1866
1867 /*********************************************************************
1868  *              _mbspbrk (MSVCRT.@)
1869  */
1870 unsigned char* CDECL _mbspbrk(const unsigned char* str, const unsigned char* accept)
1871 {
1872     const unsigned char* p;
1873
1874     while(*str)
1875     {
1876         for(p = accept; *p; p += (_ismbblead(*p)?2:1) )
1877         {
1878             if (*p == *str)
1879                 if( !_ismbblead(*p) || ( *(p+1) == *(str+1) ) )
1880                      return (unsigned char*)str;
1881         }
1882         str += (_ismbblead(*str)?2:1);
1883     }
1884     return NULL;
1885 }
1886
1887
1888 /*
1889  * Functions depending on locale codepage
1890  */
1891
1892 /*********************************************************************
1893  *              mblen(MSVCRT.@)
1894  * REMARKS
1895  *  Unlike most of the multibyte string functions this function uses
1896  *  the locale codepage, not the codepage set by _setmbcp
1897  */
1898 int CDECL MSVCRT_mblen(const char* str, MSVCRT_size_t size)
1899 {
1900   if (str && *str && size)
1901   {
1902     if(get_locinfo()->mb_cur_max == 1)
1903       return 1; /* ASCII CP */
1904
1905     return !MSVCRT_isleadbyte(*str) ? 1 : (size>1 ? 2 : -1);
1906   }
1907   return 0;
1908 }
1909
1910 /*********************************************************************
1911  *              _mbstrlen_l(MSVCRT.@)
1912  */
1913 MSVCRT_size_t CDECL _mbstrlen_l(const char* str, MSVCRT__locale_t locale)
1914 {
1915     MSVCRT_pthreadlocinfo locinfo;
1916
1917     if(!locale)
1918         locinfo = get_locinfo();
1919     else
1920         locinfo = locale->locinfo;
1921
1922     if(locinfo->mb_cur_max > 1) {
1923         MSVCRT_size_t len;
1924         len = MultiByteToWideChar(locinfo->lc_codepage, MB_ERR_INVALID_CHARS,
1925                                   str, -1, NULL, 0);
1926         if (!len) {
1927             *MSVCRT__errno() = MSVCRT_EILSEQ;
1928             return -1;
1929         }
1930         return len - 1;
1931     }
1932
1933     return strlen(str);
1934 }
1935
1936 /*********************************************************************
1937  *              _mbstrlen(MSVCRT.@)
1938  */
1939 MSVCRT_size_t CDECL _mbstrlen(const char* str)
1940 {
1941     return _mbstrlen_l(str, NULL);
1942 }
1943
1944 /*********************************************************************
1945  *              _mbtowc_l(MSVCRT.@)
1946  */
1947 int CDECL MSVCRT_mbtowc_l(MSVCRT_wchar_t *dst, const char* str, MSVCRT_size_t n, MSVCRT__locale_t locale)
1948 {
1949     MSVCRT_pthreadlocinfo locinfo;
1950     MSVCRT_wchar_t tmpdst = '\0';
1951
1952     if(!locale)
1953         locinfo = get_locinfo();
1954     else
1955         locinfo = locale->locinfo;
1956
1957     if(n <= 0 || !str)
1958         return 0;
1959     if(!MultiByteToWideChar(locinfo->lc_codepage, 0, str, n, &tmpdst, 1))
1960         return -1;
1961     if(dst)
1962         *dst = tmpdst;
1963     /* return the number of bytes from src that have been used */
1964     if(!*str)
1965         return 0;
1966     if(n >= 2 && MSVCRT__isleadbyte_l(*str, locale) && str[1])
1967         return 2;
1968     return 1;
1969 }
1970
1971 /*********************************************************************
1972  *              mbtowc(MSVCRT.@)
1973  */
1974 int CDECL MSVCRT_mbtowc(MSVCRT_wchar_t *dst, const char* str, MSVCRT_size_t n)
1975 {
1976     return MSVCRT_mbtowc_l(dst, str, n, NULL);
1977 }
1978
1979 /*********************************************************************
1980  *              _mbstowcs_l(MSVCRT.@)
1981  */
1982 MSVCRT_size_t CDECL MSVCRT__mbstowcs_l(MSVCRT_wchar_t *wcstr, const char *mbstr,
1983         MSVCRT_size_t count, MSVCRT__locale_t locale)
1984 {
1985     MSVCRT_pthreadlocinfo locinfo;
1986     MSVCRT_size_t i, size;
1987
1988     if(!locale)
1989         locinfo = get_locinfo();
1990     else
1991         locinfo = locale->locinfo;
1992
1993     /* Ignore count parameter */
1994     if(!wcstr)
1995         return MultiByteToWideChar(locinfo->lc_codepage, 0, mbstr, -1, NULL, 0)-1;
1996
1997     for(i=0, size=0; i<count; i++) {
1998         if(mbstr[size] == '\0')
1999             break;
2000
2001         size += (MSVCRT__isleadbyte_l(mbstr[size], locale) ? 2 : 1);
2002     }
2003
2004     size = MultiByteToWideChar(locinfo->lc_codepage, 0,
2005             mbstr, size, wcstr, count);
2006
2007     if(size<count && wcstr)
2008         wcstr[size] = '\0';
2009
2010     return size;
2011 }
2012
2013 /*********************************************************************
2014  *              mbstowcs(MSVCRT.@)
2015  */
2016 MSVCRT_size_t CDECL MSVCRT_mbstowcs(MSVCRT_wchar_t *wcstr,
2017         const char *mbstr, MSVCRT_size_t count)
2018 {
2019     return MSVCRT__mbstowcs_l(wcstr, mbstr, count, NULL);
2020 }
2021
2022 /*********************************************************************
2023  *              _mbstowcs_s_l(MSVCRT.@)
2024  */
2025 int CDECL MSVCRT__mbstowcs_s_l(MSVCRT_size_t *ret, MSVCRT_wchar_t *wcstr,
2026         MSVCRT_size_t size, const char *mbstr,
2027         MSVCRT_size_t count, MSVCRT__locale_t locale)
2028 {
2029     MSVCRT_size_t conv;
2030
2031     if(!wcstr && !size) {
2032         conv = MSVCRT__mbstowcs_l(NULL, mbstr, 0, locale);
2033         if(ret)
2034             *ret = conv+1;
2035         return 0;
2036     }
2037
2038     if (!MSVCRT_CHECK_PMT(wcstr != NULL)) return MSVCRT_EINVAL;
2039     if (!MSVCRT_CHECK_PMT(mbstr != NULL)) {
2040         if(size) wcstr[0] = '\0';
2041         return MSVCRT_EINVAL;
2042     }
2043
2044     if(count==MSVCRT__TRUNCATE || size<count)
2045         conv = size;
2046     else
2047         conv = count;
2048
2049     conv = MSVCRT__mbstowcs_l(wcstr, mbstr, conv, locale);
2050     if(conv<size)
2051         wcstr[conv++] = '\0';
2052     else if(conv==size && (count==MSVCRT__TRUNCATE || wcstr[conv-1]=='\0'))
2053         wcstr[conv-1] = '\0';
2054     else {
2055         MSVCRT_INVALID_PMT("wcstr[size] is too small", MSVCRT_ERANGE);
2056         if(size)
2057             wcstr[0] = '\0';
2058         return MSVCRT_ERANGE;
2059     }
2060
2061     if(ret)
2062         *ret = conv;
2063     return 0;
2064 }
2065
2066 /*********************************************************************
2067  *              mbstowcs_s(MSVCRT.@)
2068  */
2069 int CDECL MSVCRT__mbstowcs_s(MSVCRT_size_t *ret, MSVCRT_wchar_t *wcstr,
2070         MSVCRT_size_t size, const char *mbstr, MSVCRT_size_t count)
2071 {
2072     return MSVCRT__mbstowcs_s_l(ret, wcstr, size, mbstr, count, NULL);
2073 }